Predictors of mortality in the heart failure and chronic kidney disease co-morbid cohort
N Chatterjee, H Sanathkumar, D BanerjeeAbstract
Abstract
Introduction and Purpose: Patients with chronic kidney disease (CKD) and concomitant heart failure (HF) have a much higher mortality rate than patients with either condition alone. Despite this, there is very little research specifically directed towards this cohort, either to identify biomarkers for prognostic purposes or to develop new treatment pathways. The novel combined CKD-HF clinic at a UK teaching hospital was set up to personalise treatment specifically for this co-morbid population and to understand how we can better help the cohort. This study uses data from the clinic to identify physical and biochemical parameters that are associated with increased mortality in the group.
Methods
Physical and biochemical parameters from each patient’s first and most recent appointments at the CKD-HF clinic were extracted and analysed using multiple logistic and Cox proportional hazards regression modelling to identify those significantly associated with mortality or reduced survival duration.
Results
753 patients were identified with a median age of 77.75 years (IQR: 67.15-83.55) and a median time of 365 days (IQR: 252-581) between first and most recent appointments. Our first (n=275) and second (n=296) models only included data from patients’ first appointments and showed that higher age and serum NT-proBNP were associated with higher mortality at 1-year or any time-point after clinic admission, whereas higher serum albumin had the opposite effect. When the most recent clinic appointment data were included, the third model (n=215) showed the same impact of age however, it was the most recent NT-proBNP associated with worsening mortality and the most recent serum albumin which was protective. This model was the best at predicting mortality with an AUC of 0.80. Age at admission and serum NT-proBNP were consistently associated with an increased odds of mortality by 4.5-6.1% per-year-added and 0.004-0.009% per-unit-added respectively. Serum albumin was the only negatively associated parameter and decreased the odds of mortality by 6.5-8.1% per-unit-increase. Survival analyses (n=296) reflected these results and showed that an age between 80-100 (p<0.0001), serum albumin below 30g/dL (p<0.01), and serum NT-proBNP above 10,000 ng/L (p<0.0001) individually had the most negative impacts on survival duration.
Conclusions
The models and survival curves produced in this study could be used for prognostication in the CKD-HF patient population. Identification of these mortality-associated parameters may also provide new therapeutic avenues resulting in better clinical outcomes one of which could be by increasing serum albumin (or reducing loss) more aggressively given that higher recent serum albumin levels are associated with reduced mortality. Lastly, the joint CKD-HF clinic model can be implemented in other hospitals allowing more accurate modelling of mortality whilst also providing a higher standard of care for the co-morbid population.